Classification of fMRI Data in the NeuCube Evolving Spiking Neural Network Architecture

This paper presents a new method and a case study on fMRI spatio- and spectro-temporal data (SSTD) classification with the use of the recently proposed NeuCube architecture [1]. NeuCube is a three dimensional brain-like model of evolving spiking neurons that can be trained with SSTD such as fMRI, EEG and other brain data. This SSTD is mapped, analyzed, modeled and trained, and the result from these processes can be used to better understand the brain processes and to better recognize brain patterns, and thus to extract new knowledge that may reside within the SSTD. From the experimental results we can conclude that the NeuCube architecture is capable of producing significantly more accurate classification results when compared with standard machine learning methods such as SVM and MLP. Moreover, the NeuCube method facilitates deep learning of the SSTD and deeper analysis of the spatio-temporal characteristics and patterns in the fMRI SSTD.